Wildfire Risk Metric Impact on Public Safety Power Shut-off Cost Savings

Abstract

Public Safety Power Shutoffs (PSPS) are a proactive strategy to mitigate fire hazards from power system infrastructure failures. System operators employ PSPS to deactivate portions of the electric grid with heightened wildfire risks to prevent wildfire ignition and redispatch generators to minimize load shedding. A measure of vegetation flammability, called the Wildland Fire Potential Index (WFPI), has been widely used to evaluate the risk of nearby wildfires to power system operation. However, the WFPI does not correlate as strongly to historically observed wildfire ignition probabilities (OWIP) as WFPI-based the Large Fire Probability (WLFP).Prior work chose not to incorporate wildfire-driven failure probabilities, such as the WLFP, because constraints with Bernoulli random variables to represent wildfire ignitions could require non-linear or non-convex constraints. This paper uses a deterministic equivalent of an otherwise complicating line de-energization constraint by quantifying the wildfire risk of operating transmission line as a sum of each energized line's wildfire ignition log probability (log(WIP)) rather than as a sum of each energized line's WFPI. A day-ahead unit commitment and line de-energization PSPS framework is used to assess the cost differences driven by the choice between the WFPI and WLFP risk metrics. Training the optimization on scenarios developed by mapping WLFP to log(WIP) rather than mapping the WFPI to log(WIP) leads to reductions in the total real-time costs. For the IEEE RTS 24-bus test system, mapping transmission line WLFP values to log(WIP) resulted in a 14.8 % (on average) decrease in expected real-time costs.

Figures (17)

• Figure 1: Block diagram showing the data inputs and decision outputs for each stage of the day-ahead and real-time optimization.
• Figure 2: A schematic of the IEEE RTS 24-bus system with each transmission line and bus highlighted to depict its WFPI-based large fire probability (WLFP) for July 31st, 2020. Note the geographic layout comes from RTSGMLC.
• Figure 3: A comparison of time series of the 7-day moving average of WFPI (top) and WLFP (bottom) for node 23 on the IEEE RTS 24-bus system near San Bernadino WFPI in 2015. The transparent red region indicates the time-span of the largest fire in San Bernadino in 2015, the Lake Fire.
• Figure 4: A schematic of the reduced 240-bus reduced WECC system WECC
• Figure 5: A comparison of time series of the WIP predicted from WFPI (left) and WLFP (right) near each bus and along each of the transmission lines for the 24-bus Reliability Test System in 2020. Dates of elevated levels of WIP predicted from WLFP correspond more closely to increases in the historically observed large wildfire ignitions in Southern California than from WIP predicted from WFPI.